RESEARCH Open Access

ESCRT-III subunits Snf7-1 and Snf7-2 differentiallyregulate transmembrane cargos in hESC-derivedhuman neuronsJin-A Lee1*, Lei Liu2, Robyn Javier2, Anatol C Kreitzer2, Celine Delaloy2 and Fen-Biao Gao3*

Abstract

Backgrounds: Endosomal sorting complex required for transport (ESCRT) is involved in several fundamentalcellular processes and human diseases. Many mammalian ESCRT proteins have multiple isoforms but their precisefunctions remain largely unknown, especially in human neurons.

Results: In this study, we differentiated human embryonic stem cells (hESCs) into postmitotic neurons andcharacterized the functional properties of these neurons. Moreover, we found that among the three humanparalogs of the yeast ESCRT-III subunit Snf7, hSnf7-1 and hSnf7-2 are most abundantly expressed in humanneurons. Both hSnf7-1 and hSnf7-2 are required for the survival of human neurons, indicating a non-redundantessential function. Indeed, hSnf7-1 and hSnf7-2 are preferentially associated with CHMP2A and CHMP2B,respectively, and regulate the turnover of distinct transmembrane cargos such as neurotransmitter receptors inhuman neurons.

Conclusion: These findings indicate that different mammalian paralogs of the yeast ESCRT-III subunit Snf7 havenon-redundant functions in human neurons, suggesting that ESCRT-III with distinct subunit compositions maypreferentially regulate different cargo proteins.

Keywords: human embryonic stem cells, neurons, ESCRT, CHMP2B, Snf7

BackgroundEndosomal sorting complex required for transportESCRTs (ESCRT-0, ESCRT-I, ESCRT-II, and ESCRT-III) are evolutionarily conserved multiunit proteins thatare not only required for many fundamental cellularprocesses, such as multivesicular body (MVB) formation,cytokinesis, virus budding, and autophagy [1-3]. In yeast,ESCRT-III is composed of four subunits, Vps20, Snf7,Vps2, and Vps24, and the Vps20-Snf7 subcomplex isrecruited from the cytoplasm to the endosomal mem-brane and then bound by the Vps2-Vps24 subcomplex[4]. In contrast, there are three human paralogs of Snf7(hSnf7-1/CHMP4A, hSnf7-2/CHMP4B, and hSnf7-3/CHMP4C) and two paralogs of Vps2 (CHMP2A and

CHMP2B) [3]. However, the functional differencebetween various ESCRT-III paralogs remains unknown.ESCRTs are also implicated in several human diseases

including neurodegeneration, cancer, and HIV infection[5]. For instance, a splice site mutation in CHMP2B isassociated with frontotemporal dementia linked to chro-mosome 3 (FTD3) and produces a mutant protein calledCHMP2BIntron5, which lacks the C-terminal 36 aminoacids [6]. Expression of CHMP2BIntron5 in rodent pri-mary neurons or other cell types blocks the proper dis-sociation of ESCRT-III, resulting in autophagosomeaccumulation, dendritic retraction, and eventual neuro-nal cell loss [7-10]. This neurotoxicity is likely due toenhanced association between the disease proteinCHMP2BIntron5 and Snf7-2 [7]. Among ESCRT-III com-ponents, Snf7 is thought to play a key role and form oli-gomers on endosomal membranes [11-13]. Indeed,acute knockdown of Snf7-2 but not CHMP2B in rodent

* Correspondence: leeja@hnu.kr; fen-biao.gao@umassmed.edu1Department of Biotechnology, College of Life Science and Nanotechnology,Hannam University, Dajeon 305-811, Korea3Department of Neurology, University of Massachusetts Medical School,Worcester, Massachusetts, 01605 USAFull list of author information is available at the end of the article

Lee et al. Molecular Brain 2011, 4:37http://www.molecularbrain.com/content/4/1/37

© 2011 Lee et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative CommonsAttribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction inany medium, provided the original work is properly cited.

cortical neurons leads to a rapid neuronal cell loss andSnf7-2 knockout mice are embryonic lethal [7].Oddly, Snf7-1 has not been identified in rodents and

its precise functions in neurons remain to be investi-gated. Here we characterized human postmitotic neu-rons differentiated from human embryonic stem cells(hESCs) and explored the non-redundant functions ofhSnf7-1 and hSnf7-2 in human neurons.

MethodsNeuronal differentiation of hESCshESCs were differentiated into neurons as described[14]. Briefly, hESC colonies were detached from thefeeder layer with 0.5 mg/ml dispase and cultured asaggregates in suspension for 4 days in hESC medium(Dulbecco’s modified Eagle medium [DMEM]/12 with20% knockout serum replacement) without FGF2.Cells were transferred to new culture dishes in thefirst 2 days to remove adherent MEFs. On day 5, EBswere transferred to flasks precoated with poly-L-ornithine/laminin (20 μg/ml) in neural inductionmedium (DMEM/F12/N2) consisting of 33% F12, 66%DMEM, 1X N2, 1% NEAA, 10 ng/ml FGF2, and 2 mg/ml heparin. Twelve days after neural induction, neu-roepithelial cells in rosettes were isolated from thesurrounding cells with 0.2 mg/ml dispase and trans-ferred to new flasks for 2-3 h to allow non-neuronalcells to attach. Floating cells were transferred to aflask coated with poly-HEME (Sigma) to prevent cellattachment.Terminal differentiation was induced as described [15]

with some modifications. At 3-4 weeks of age, neuro-spheres were dissociated with Accutase (Sigma), platedon glass coverslips coated with poly-D-lysine and lami-nin (BD Biosciences), and cultured in neurobasal med-ium supplemented with 2% B27, 1% nonessential aminoacids solution, 0.5 mM L-glutamine, 1 μg/ml laminin(Sigma), 10 ng/ml BDNF, and 10 ng/ml GDNF (R&DSystems) for 1-3 weeks. Half of the culture medium waschanged every other day.

ImmunocytochemistryOn day 7 or 14 after terminal differentiation, humanneurons were fixed with 4% paraformadehyde for 10min at room temperature (RT) and permeabilized with0.1% Triton × 100 for 5 min at RT. After blocking with3% BSA (Sigma), cells were incubated with primary anti-bodies [anti-nestin (1:100), anti-MAP2 (1:100), anti-BF1(1:100), anti-GFAP (1:100), anti-PSD95 (1:100)] for 1 hat RT and then with secondary antibodies (anti-mouseor anti-rabbit conjugated with Cy3 or Alexa 488) for 45min at RT. Anti-NR1 antibody recognizes the syntheticpeptide RAEPDPKKKATFRA, corresponding to aminoacids 849-862 of NMDAR1.

Gene expression analysis during neuronal differentiationof hESCsTotal RNA was isolated from human neurons at 3 or 14DIV and incubated with DNase I for 30 min at 37°C.Reverse transcription was performed with Taq-Manreagent (Applied Biosystem). qRT-PCR was performedwith an ABI Prism 7700 Thermocycler with fluorescencedetection (Applied Biosystems) as described [14]. Theprimers were Snf7-1 (sense 5’-cctatgggctttggagatga-3’,antisense 5’-ttgtcgcccacatttaacaa-3’); Snf7-2 (sense 5’-cgaaacctgtagggtttgga-3’, antisense 5’-ctgtttcgggtccactgatt-3’); Snf7-3 (sense 5’-gaactccacagcaatgagca-3’, antisense5’-ccagcatctcctcagtctcc-3’); NR2B (sense 5’-tctttgga-gatggggagatg-3’, antisense 5’-tcgcagatgaaggtgatgag-3’);mGluR1 (sense 5’-ggaagggaattatggggaga-3’, antisense 5’-tgtcatgccttcacaggagc-3’). The primers for GAPDH (GeneID: NM_00204.3) were from ABI.

Electrophysiology of human neuronsCells targeted for whole-cell patch-clamp recordingswere visualized with infrared differential interferencecontrast microscopy (Olympus BX51WI). Cells wereimmersed in a HEPES-buffered saline solution contain-ing 115 mM NaCl, 2 mM KCl, 10 mM HEPES, 3 mMCaCl2, 1.5 mM MgCl2, and 10 glucose. All experimentswere performed at RT. Resistance of micropipettes was2.5-4.0 MΩ when filled with the following: 130 mMKMeSO3, 10 mM NaCl, 2 mM MgCl2, 0.16 mM CaCl2,10 mM HEPES, and 0.5 mM EGTA. Current- and vol-tage-clamp recordings were obtained with a Multiclamp700B amplifier (Molecular Devices), filtered at 2 kHz,and digitized at 10 kHz. All data were acquired and ana-lyzed with custom programs written in Igor Pro.

Induction of depolarization of human neuronsTo induce depolarization, human neurons (14-21 DIV)were incubated in neurobasal medium (Invitrogen) with1 μM tetrodotoxin (Sigma) and 10 μM 2,3-dihydroxy-6-nitro-7-sulfamoyl-benzo[f]quinoxaline-2,3-dione (Sigma)for 30 min at 37°C. In some experiments, 2X depolariza-tion solution (10 mM HEPES pH 7.4, 1.8 mM CaCl2,110 mM KCl, 0.925 mM NaH2PO4, 0.39 mM MgSO4,5.36 mM NaCl) was added. Depolarization-induced geneexpression was determined by qRT-PCR.

Gene knockdown in human neuronsFor generation of virus, each RNA was subcloned intothe pSicoR vector (McManus lab, University of Califor-nia, San Francisco). To knock down hSnf7-1 and hSnf7-2 in human neurons, specific sequences were selected astargets of siRNA (Additional file 1). For immunoblotanalysis, proteins extracts were separated by SDS-PAGEand transferred to a PDVF membrane. After blockingwith 5% non-fat milk in 1% Tween 20 in PBS, the

Lee et al. Molecular Brain 2011, 4:37http://www.molecularbrain.com/content/4/1/37

Page 2 of 8

membrane was incubated first with primary antibody at4°C overnight and then with HRP-conjugated anti-mouse or anti-rabbit secondary antibody at RT for 1 h.

Survival assay for human neuronsTo investigate the effect of loss of hSnf7-1 or hSnf7-2 inhuman neurons, we infected mature cortical neurons(14-16 DIV) with lentivirus expressing siRNA. PI- andGFP-positive neurons were counted as surviving neu-rons. Three independent cell survival assays were done.

ResultsDifferentiation and characterization of human postmitoticneurons derived from hESCsTo establish a human cellular model of neurodegenera-tion, hESCs (H9 cells) were differentiated into postmito-tic neurons as previously reported [14]. Briefly, hESCswere grown on mouse embryonic fibroblasts (MEFs) inthe presence of fibroblast growth factor 2 (FGF2).Embryoid bodies (EBs) formed in 4 days from

dissociated hESCs in the absence of FGF2 and were dif-ferentiated in the presence of FGF2 and N2 supplementinto a monolayer containing rosette structures. Rosettecells were isolated and placed in suspension cultures toallow human neural progenitor cells (hNPCs) in neuro-spheres to proliferate for a month. hNPCs dissociatedfrom neurospheres were differentiated into postmitoticneurons and glia in the presence of brain-derived neuro-trophic factor (BDNF) and glial cell line-derived neuro-trophic factor (GDNF) (Figure 1A).To use human postmitotic neurons as a cellular model

to understand molecular pathogenesis of FTD, we usedthe protocol described above to differentiate hESCs intopostmitotic neurons with characteristics of forebrainneuronal lineages. For instance, hNPCs in rosette struc-tures express nestin and BF1, a transcription factorexpressed in telencephalic neural progenitor cells(Figure 1B). hNPCs in neurospheres also express Pax6, atranscription factor containing a paired domain and apaired-type homeodomain involved in early telencephalon

09.175A (09.168A.micro) 09.175A (09.168C.micro)

09.175A (09.168D.micro) 09.175A (09.168F.micro)

09.175A (09.168G.micro) 09.175A (09.168I.micro)

A B D

BF1

Nestin MAP2

Merged

hESCs on MEF

Embryoid Bodies(0–4 DIV)

Rosettes(4–16 DIV)

Neurospheres(16–46 DIV)

Neurons(> 46 DIV)

PSD-95

Merged

C MAP2

GFAP

Merged

Figure 1 Differentiation of human postmitotic neurons derived from hESCs. (A) Differentiation of H9 hESCs into neurons. hESCs (line H9)were dissociated from feeder cells and cultured in aggregates to develop into EBs, which were transferred to flasks containing neural inductionmedium. Rosette structures were formed and, 12 days after neural induction, were isolated from the surrounding cells to develop intoneurospheres. One month later, terminal neuronal differentiation was induced in the presence of BDNF and GDNF. (B) Human neural progenitorsat the rosette stage were immunostained with anti-nestin antibody (green) and anti-BF1 antibody (red). Bars: 50 mm. (C) Terminally differentiatedhuman neural cells were immunostained with anti-MAP2 antibody (red) and anti-GFAP antibody (green). Bars: 50 μm. (D) Human neuronsderived from hESCs were immunostained with anti-MAP2 antibody (green) and anti-PSD-95 antibody (red). The merged image shows anenlarged image of the boxed area. Bars: 20 μm.

Lee et al. Molecular Brain 2011, 4:37http://www.molecularbrain.com/content/4/1/37

Page 3 of 8

patterning [16]. After terminal differentiation, 70-80% ofcells were MAP2-positive neurons, with a few GFAP-posi-tive astrocytes (Figure 1C). PSD-95-immunoreactivepuncta were formed on dendrites of these human postmi-totic neurons (Figure 1D), suggesting the formation ofpotential synapses.These neurons expressed receptors for neurotransmit-

ters, such as glutamate receptor 1 (GluR1) and NMDAreceptor 2B (NR2B), as measured by quantitative real-time polymerase chain reaction (qRT-PCR) (Figure 2A)or immunostaining with receptor-specific antibodies(Figure 2B and 2C). In the presence of 50 mM KCl, c-fos mRNA expression increased markedly considerably(Figure 2D), indicating that depolarization at the mem-brane can be coupled with gene expression in thenucleus of these neurons.These human postmitotic neurons displayed electro-

physiological properties similar to those of rodent neu-rons in primary cultures. In voltage-clamp studies, aseries of depolarizing voltage steps elicited an inward Na+ current that rapidly inactivated (Figure 2E) and alonger sustained outward current mediated predomi-nately by K+ (Figure 2F). Current-voltage plots for peakNa+ and K+ currents revealed voltage-sensitive ion chan-nels activating in the range of -40 to -20 mV. To

confirm that these membrane conductances supportaction potential generation, we performed current-clamp recordings (Figure 2G). Neurons exhibited nega-tive resting membrane potentials in the physiologicalrange (-57.5 ± 2.2 mV, n = 4), and action potentialswere observed in all four cells tested. For the humanneurons, positive current injection elicited an actionpotential (see inset) followed by sustained depolarizationand inactivation of further spiking (Figure 2H). Theseresults indicate that these hESCs-derived human postmi-totic neurons have molecular and electrophysiologicalcharacteristics similar to those of native neurons.

hSnf7-1 and hSnf7-2 are expressed in human postmitoticneuronsSnf7 is a key component of yeast ESCRT-III [1]. OneSnf7 homolog (Shrub) is found in Drosophila, and two(mSnf7-2/CHMP4B and mSnf7-3/CHMP4C) are foundin mice [17]. Enhanced association was observedbetween mSnf7-2 and CHMP2BIntron5, a mutant pro-tein implicated in FTD3 [7]. In humans, three Snf7homologs have been reported (hSnf7-1/CHMP4A,hSnf7-2/CHMP4B, and hSnf7-3/CHMP4C)[3]. How-ever, the mouse counterpart of hSnf7-1/CHMP4A hasnot been identified. Since the rodent homolog for

09.175A (09.177A.line)09.175A (09.177B.line)

09.175A (09.177C.line)09.175A (09.177D.line)

09.175A (09.168J.micro) 09.175A (09.168K.micro)

A

E F G H

14 DIV Neurons3 DIV Neurons

INa

INa

0.2 nA2 ms

0.2 nA50 ms

20 mV

–55 mV

10 ms100 ms–80 –40

–1.6

0.8

nA1.6

–0.840 80

IKIK

01

3

Rel

ativ

e G

ene

Exp

ress

ion

NR2BGluR1

mV

5

7

9

11

13

GluR1 NR2B

50 mM KCl, 4 hrcontrol

0

1.0

2.0

Rel

ativ

e Fo

s Le

vel **

3.0

B C D

Figure 2 Characterization of human neurons derived from hESCs. (A) Expression of NR2B or GluR1 in human postmitotic neurons wasconfirmed by qRT-PCR analysis. Values are mean ± s.e.m. (B) Immunostaining of neurons at 14 DIV with anti-GluR1 antibody. Bar: 20 μm. (C)Immunostaining of neurons at 14 DIV with anti-NR2B antibody. Bars: 20 μm. (D) Membrane depolarization by 50 mM KCl treatment increased c-fos expression in hESC-derived human neurons as measured by RT-PCR analysis. Values are mean ± s.e.m. (E-H) Electrophysiological properties ofhuman neurons derived from hESCs. Voltage-sensitive sodium and potassium currents measured in voltage-clamp (E-G) support the generationof action potentials in response to current injection (H).

Lee et al. Molecular Brain 2011, 4:37http://www.molecularbrain.com/content/4/1/37

Page 4 of 8

hSnf7-1 has not been identified, the roles of hSnf7-1and hSnf7-2 in neurodegeneration could only be stu-died in human neurons. To characterize the functionsand contributions of Snf7 family proteins to neurode-generation in hESCs-derived human neurons, we firstcompared the relative expression levels of human Snf7family genes by qRT-PCR in human neurons culturedfor 7 days in vitro (DIV). hSnf7-2 mRNA was threetimes more abundant than hSnf7-1, but the amount ofhSnf7-3 mRNA was <0.1% of hSnf7-1 mRNA (Figure3A). Thus, hSnf7-1 and hSnf7-2 are major Snf7 familyproteins expressed in human postmitotic neurons.Immunostaining revealed hSnf7-2 in the cytoplasm inhESCs (Figure 3B). Both hSnf7-1 and hSnf7-2 wereexpressed in human postmitotic neurons and seemedto be present in both the nucleus and cytoplasm sug-gesting that they must have some roles in human neu-rons (Figure 3D, E).

Both hSnf7-1 and hSnf7-2 are essential for the survival ofhuman postmitotic neuronsTo dissect the functions of hSnf7-1 and hSnf7-2 inhuman postmitotic neurons, we generated hSnf7-1- orhSnf7-2-specific siRNAs and checked efficiency of theirknockdown in HEK293 cells (Table S1). We chose onesiRNA with most strong knockdown-effect among siR-NAs of hSnf7-1 or hSnf7-2 and generated lentivirusconstructs expressing hSnf7-1- or hSnf7-2-specific siR-NAs. At 21 DIV, fully differentiated human mature

neurons were transfected with hSnf7-1 siRNA lentivirus(#2) or hSnf7-2 siRNA lentivirus (#6). Each siRNA speci-fically reduced gene expression of each paralog inhuman neurons (Figure 4A). Interestingly, reducedexpression of hSnf7-1 or hSnf7-2 caused significant neu-ronal cell loss by days 3 and 4 after infection, suggestingnon-redundant functions for this family of ESCRT-IIIcomponents (Figure 4B). This result implicates thatboth hSnf7-1 and hSnf7-2 are required for human neu-ronal cell survival and exert non-redundant functions.Consistent with earlier reports on mSnf7-2 [6,7], loss

of hSnf7-2 in human neurons also led to accumulationof autophagosomes, as indicated by GFP-LC3 puncta, aphenotype similar to that caused by the expression ofthe FTD3-associated mutant protein CHMP2BIntron5

(Figure 5B, C). Accordingly, loss of hSnf7-1 by siRNAcaused a similar cellular phenotype in human neurons(Figure 5D), again indicating that hSnf7-1 and hSnf7-2play non-redundant roles in ESCRT-III related cellularprocesses.

hSnf7-1- or hSnf7-2-containing ESCRT-III differentiallyregulates the turnover of distinct transmembrane cargosin human neuronsTo further explore the functional differences betweenhSnf7-1 and hSnf7-2, we determined whether hSnf7-1and hSnf7-2 are present in the same ESCRT-III. Amongthe components of yeast ESCRT-III, only Vps2 hasmore than one human homolog (CHMP2A and

A

0

1

2

3

4

Rel

ativ

e m

RN

A L

evel

s

hSnf7-1

hSnf7-2

hSnf7-3

CB

MAP2

MAP2

hSnf7-1 Merged

MergedhSnf7-2

D

E

Figure 3 hSnf7-1 and hSnf7-2 are expressed in human post-mitotic neurons. (A) qRT-PCR showed that hSnf7-2 was highly expressed inhuman neurons, while hSnf7-3 was barely detectable. (B) hESCs were immunostained with hSnf7-2. hSnf7-2 was mostly localized to cytoplasm.(C) Nucleus staining (indicated by DAPI staining) of hESCs as shown in panel B. (D) Immunostaining showed that hSnf7-1 was mostly expressedin the cytoplasm of human neurons. Bar: 20 mm. (E) Immunostaining with a hSnf7-2-specific antibody reveals hSnf7-2 in both the nucleus andcytoplasm of human neurons. Bar: 20 μm.

Lee et al. Molecular Brain 2011, 4:37http://www.molecularbrain.com/content/4/1/37

Page 5 of 8

CHMP2B); vps20 and vps24 have only one each (CHMP6and CHMP3) [3,18,19]. Therefore, we examined the bio-chemical interactions between human homologs of Snf7and Vps2 family proteins. Flag-CHMP2A or Flag-CHMP2B was expressed in HEK293T cells. Two dayslater, immunoprecipitation (IP) was performed with anti-Flag antibody, and immunoisolates were analyzed by wes-tern blot with antibodies against hSnf7-1 or hSnf7-2.Intriguingly, endogenous hSnf7-1 preferentially

associated with CHMP2A, and complexes containinghSnf7-2 and CHMP2B were more abundant than thosecontaining CHMP2A (Figure 6A). Thus, hSnf7-1 andhSnf7-2 may have preferred interacting partners to formdistinct ESCRT-III with different cellular functions.To test this hypothesis, we examined whether hSnf7-

1- or hSnf7-2-contaning ESCRT-III regulates the traf-ficking of different transmembrane cargos in the MVBpathway. To this end, we examined the degradation andaccumulation of different transmembrane cargos as aresult of loss of hSnf7-1 or hSnf7-2 activity in maturehESC-derived human postmitotic neurons. The cellswere infected at 21-23 DIV with lentivirus expressinghSnf7-1 or hSnf7-2 siRNA. These siRNAs specificallyreduced endogenous hSnf7-1 or hSnf7-2 in human neu-rons (Figure 4A).We then examined the protein levels of some recep-

tors essential for neuronal cell survival and signaling inhuman neurons 2-3 days after lentiviral infections.Reduced expression of hSnf7-1 or hSnf7-2 led to accu-mulation of epidermal growth factor receptor (EGFR)but did not affect the level of GluR1 (Figure 6B). Inter-estingly, the nerve growth factor (NGF) receptor TrkAaccumulated to a greater extent after knockdown ofhSnf7-1 than of hSnf2. In contrast, amyloid precursorprotein (APP), a transmembrane protein implicated inAlzheimer’s disease, and NMDA receptor 1 (NR1) accu-mulated more after knockdown of hSnf7-2 (Figure 6B).

DiscussionOur findings as reported here suggest that differentESCRT-III subunits may associate with each other pre-ferentially to form ESCRT-III with unique subunit com-positions and distinct cellular functions in humanneurons.

Days After Transfection

020406080

100

Cel

l Su

rviv

al (

%)

1 2 3

hSnf7-1 siRNAScrambled

***

hSnf7-2 siRNA

***

*

***

***

***

4

B

hSnf7-1

hSnf7-2

Actin

Actin

Cont. #1 #2 siRNA

#5 #6 siRNACont.

A

Figure 4 Both hSnf7-1 and hSnf7-2 are essential for the survival of human neurons. Terminally differentiated human neurons (14-16 DIV)were infected with lentivirus expressing either hSnf7-1 or hSnf7-2 siRNA. (A) Western blot analysis showed hSnf7-1 and hSnf7-2 specifically theirexpression. (B) Infection of lentivirus expressing either hSnf7-1 or hSnf7-2 siRNA caused human neuronal cell loss, respectively. GFP-positive andPI-negative neurons were considered as surviving neurons.

CHMP2BIntron5

hSnf7-1 siRNAhSnf7-2 siRNA

CHMP2BWT

A B

C D

Figure 5 Autophagosome formation in hESC-derived humanneurons. (A) A CHMP2BWT-transfected human neuron. (B)Autophagosomes accumulate in CHMP2BIntron5-transfected humanneurons, as indicated by GFP-LC3. (C) SiRNA knockdown of hSnf7-2in human neurons leads to autophagosome accumulation, asindicated by GFP-LC3. (D) SiRNA knockdown of hSnf7-1 in humanneurons leads to autophagosome accumulation, as indicated byGFP-LC3.

Lee et al. Molecular Brain 2011, 4:37http://www.molecularbrain.com/content/4/1/37

Page 6 of 8

ESCRT machinery is composed of four heteromericprotein complexes, ESCRT-0, -I, -II, and -III [1-3].Intriguingly, in contrast to yeast ESCRT subunits,many mammalian ESCRT proteins have multiple iso-forms such as Snf7-1/Snf7-2/Snf7-3 (CHMP4A/CHMP4B/CHMP4C), CHMP2A/CHMP2B, andVPS37A/VPS37B/VPS37C/VPS37D [1-3,12]. Althoughthey are highly conserved, the significance and uniquefunctions of individual isoforms are largely unknown.As shown in our study, hSnf7-1 and hSnf7-2 but nothSnf7-3 are mostly expressed in human postmitoticneurons and our loss of function studies revealed theirfunctional non-redundancy. The preferential associa-tion of hSnf7-1 with CHMP2A and hSnf7-2 withCHMP2B supports the idea that multiple isoforms ofESCRT proteins may form different ESCRT complexes.Indeed, a recent study by Stefani et al. [20] showedthat VPS37A but not VPS37C formed endosome-speci-fic ESCRT-I complex with TSG101 and UBAP1. Itremains to be determined what are the exact composi-tions of distinct ESCRT-III complexes and how differ-ent transmembrane cargos are differentially recognized

in different mammalian cell types and various cellularprocesses.In our study, we used the protocol to differentiate

hESCs into human postmitotic neurons with character-istics of forebrain neuronal lineages [14-16]. Differen-tiated human postmitotic neurons have some similarmorphological and electrophysiological properties withthose of rodent neurons as shown in Figure 2. Thus,this can be a valuable model system to examine human-specific gene functions in neurons and to investigate themolecular pathogenic mechanisms of several neurode-generative diseases such as FTD.

ConclusionsIn this study, we investigated the molecular and cellularfunctions of Snf7 family proteins using hESC-derivedhuman neurons. hSnf7-1 and hSnf7-2 are most abun-dantly expressed in human neurons and both arerequired for neuronal survival, suggesting non-redun-dant cellular functions. Indeed, these two paralogs pre-ferentially form complexes with CHMP2A andCHMP2B, respectively, and seem to differentially

hSnf7-1

hSnf7-2

Flag-CHMP2A

Flag-CHMP2BControl

Flag-CHMP2A

Flag-CHMP2BControl

Flag

Input

IP with Flag Ab

hSnf7-1

hSnf7-2

Flag

A

(2.2:1.0)

(1.0:3.5)

hSnf7-1

siRNA

Scramble

d

hSnf7-2

siRNA

hSnf7-1

hSnf7-2

EGFR

TrkA

NR1

GluR1

Actin

APP

(1.0:1.8:2.4)

(1.0:3.2:1.8)

(1.0:1.1:1.2)

(1.0:1.2:2.9)

(1.0:1.7:5.3)

B

Figure 6 Differential regulation of diverse transmembrane cargos in human neurons by ESCRT-III with different subunit compositions.(A) Preferential interactions between hSnf7-1 and CHMP2A and between hSnf7-2 and CHMP2B. Flag-tagged CHMP2A or CHMP2B wastransfected into HEK293T cells, and endogenous hSnf7-1 and hSnf7-2 proteins were detected by western blot with specific antibodies after IPwith anti-Flag antibody. (B) siRNA knockdown of hSnf7-1 or hSnf7-2 in human neurons had differential effects on the accumulation of differenttransmembrane cargos. Human neurons were collected 3 days after infection with lentivirus expressing hSnf7-1 or hSnf7-2 siRNA. Actin was usedas the loading control.

Lee et al. Molecular Brain 2011, 4:37http://www.molecularbrain.com/content/4/1/37

Page 7 of 8

regulate the turnover of various transmembrane cargos.Thus, unlike in yeast, different ESCRT-III complexeswith unique subunit compositions are likely present inhuman neurons and fulfill specific cellular functions.

Additional material

Additional file 1: Target sequences in hSnf7-1 and hSn7-2 for RNAi.This file contains primer sequences used to generate hsnf7-1 and hsnf7-2-specific RNAi constructs.

AcknowledgementsThis work was initiated and performed mostly at the Gladstone Institute ofNeurological Disease in San Francisco when J.A.L. and F.-B.G. were still there.We thank P. Hanson, M.T. McManus, and L.F. Reichardt for reagents. We alsothank S. Ordway for editorial assistance, A. Wilson for help with graphics,and Gao lab members for discussions. J.-A.L. and C.D. were supported byfellowships from the California Institute for Regenerative Medicine (CIRM).This work was supported by grants from CIRM (F.-B.G.) and the NIH(RO1NS057553, F.-B.G.).

Author details1Department of Biotechnology, College of Life Science and Nanotechnology,Hannam University, Dajeon 305-811, Korea. 2Gladstone Institute ofNeurological Disease, San Francisco, California, 94158 USA. 3Department ofNeurology, University of Massachusetts Medical School, Worcester,Massachusetts, 01605 USA.

Authors’ contributionsJAL and FBG designed the experiments, analyzed the data, and wrote themanuscript. JAL and LL performed most of the experiments. RJ and AKperformed the electrophysiological experiments. C.D. was involved inestablishing hESC cultures. All authors read and approved the finalmanuscript.

Competing interestsThe author declares that they have no competing interests.

Received: 7 September 2011 Accepted: 5 October 2011Published: 5 October 2011

References1. Henne WM, Buchkovich NJ, Emr SD: The ESCRT pathway. Dev Cell 2011,

21:77-91.2. Hurley JH, Hanson PI: Membrane budding and scission by the ESCRT

machinery: it’s all in the neck. Nat Rev Mol Cell Biol 2010, 11:556-566.3. Raiborg C, Stenmark H: The ESCRT machinery in endosomal sorting of

ubiquitylated membrane proteins. Nature 2009, 458:445-452.4. Babst M, Katzmann DJ, Estepa-Sabal EJ, Meerloo T, Emr SD: Escrt-III: an

endosome-associated heterooligomeric protein complex required formvb sorting. Dev Cell 2002, 3:271-282.

5. Saksena S, Emr SD: ESCRTs and human disease. Biochem Soc Trans 2009,37:167-172.

6. Skibinski G, Parkinson NJ, Brown JM, Chakrabarti L, Lloyd SL, Hummerich H,Nielsen JE, Hodges JR, Spillantini MG, Thusgaard T, Brandner S, Brun A,Rossor MN, Gade A, Johannsen P, Sørensen SA, Gydesen S, Fisher EM,Collinge J: Mutations in the endosomal ESCRTIII-complex subunitCHMP2B in frontotemporal dementia. Nat Genet 2005, 37:806-808.

7. Lee JA, Beigneux A, Ahmad ST, Young SG, Gao FB: ESCRT-III dysfunctioncauses autophagosome accumulation and neurodegeneration. Curr Biol2007, 17:1561-1567.

8. Filimonenko M, Stuffers S, Raiborg C, Yamamoto A, Malerød L, Fisher EM,Isaacs A, Brech A, Stenmark H, Simonsen A: Functional multivesicularbodies are required for autophagic clearance of protein aggregatesassociated with neurodegenerative disease. J Cell Biol 2007, 179:485-500.

9. Lee JA, Gao FB: Inhibition of autophagy induction delays neuronal cellloss caused by dysfunctional ESCRT-III in frontotemporal dementia. JNeurosci 2009, 29:8506-8511.

10. Urwin H, Authier A, Nielsen JE, Metcalf D, Powell C, Froud K, Malcolm DS,Holm I, Johannsen P, Brown J, Fisher EM, van der Zee J, Bruyland M, FReJAConsortium, Van Broeckhoven C, Collinge J, Brandner S, Futter C, Isaacs AM:Disruption of endocytic trafficking in frontotemporal dementia withCHMP2B mutations. Hum Mol Genet 2010, 19:2228-2238.

11. Hanson PI, Roth R, Lin Y, Heuser JE: Plasma membrane deformation bycircular arrays of ESCRT-III protein filaments. J Cell Biol 2008, 180:389-402.

12. Hanson PI, Shim S, Merrill SA: Cell biology of the ESCRT machinery. CurrOpin Cell Biol 2009, 21:568-574.

13. Teis D, Saksena S, Judson BL, Emr SD: ESCRT-II coordinates the assemblyof ESCRT-III filaments for cargo sorting and multivesicular body vesicleformation. EMBO J 2010, 29:871-883.

14. Delaloy C, Liu L, Lee JA, Su H, Shen F, Yang GY, Young WL, Ivey KN, Gao FB:MicroRNA-9 coordinates proliferation and migration of humanembryonic stem cells-derived neural progenitors. Cell Stem Cell 2010,6:323-335.

15. Schneider BL, Seehus CR, Capowski EE, Aebischer P, Zhang SC,Svendsen CN: Over-expression of alpha-synuclein in human neuralprogenitors leads to specific changes in fate and differentiation. HumMol Genet 2007, 16:651-666.

16. Hébert JM, Fishell G: The genetics of early telencephalon patterning:some assembly required. Nat Rev Neurosci 2008, 9:678-685.

17. Sweeney NT, Brenman JE, Jan YN, Gao FB: The coiled-coil protein Shrubcontrols neuronal morphogenesis in Drosophila. Curr Biol 2006,16:1006-1011.

18. Peck JW, Bowden ET, Burbelo PD: Structure and function of human Vps20and Snf7 proteins. Biochem J 2004, 377:693-700.

19. Katoh K, Shibata H, Hatta K, Maki M: CHMP4b is a major binding partnerof the ALG-2-interacting protein Alix among the three CHMP4 isoforms.Arch Biochem Biophys 2004, 421:159-165.

20. Stefani F, Zhang L, Taylor S, Donovan J, Rollinson S, Doyotte A, Brownhill K,Bennion J, Pickering-Brown S, Woodman P: UBAP1 Is a Component of anEndosome-Specific ESCRT-I Complex that Is Essential for MVB Sorting.Curr Biol 2011, 21:1245-1250.

doi:10.1186/1756-6606-4-37Cite this article as: Lee et al.: ESCRT-III subunits Snf7-1 and Snf7-2differentially regulate transmembrane cargos in hESC-derived humanneurons. Molecular Brain 2011 4:37.

Submit your next manuscript to BioMed Centraland take full advantage of:

• Convenient online submission

• Thorough peer review

• No space constraints or color figure charges

• Immediate publication on acceptance

• Inclusion in PubMed, CAS, Scopus and Google Scholar

• Research which is freely available for redistribution

Submit your manuscript at www.biomedcentral.com/submit

Lee et al. Molecular Brain 2011, 4:37http://www.molecularbrain.com/content/4/1/37

Page 8 of 8